The Journal of Neuroscience, August 1993, f3(8): 31933210

The Retinal Fate of Xenopus Cleavage Stage Progenitors Is Dependent upon Blastomere Position and Competence: Studies of Normal and Regulated Clones

Sen Huang and Sally A. Moody DeDartment of Anatomv and Neuroscience Proaram, The George Washington University Medical Center, Washington, D.6. 20037

The clonal origin of the stage 43-44 Xenopus retina from neighbors. This result suggests that dorsal-animal blasto- cleavage stage precursors was quantitatively assessed with meres change positional values in only one direction (dorsal lineage tracing techniques. The retina descends from a spe- to vegetal) after neighbor cell deletion, and that retinal fate cific subset of those blastomeres that form forebrain. The is dictated by blastomere position. To test this hypothesis most animal dorsal midline cell (Dl .l. 1) produced about half directly, different ventral and vegetal blastomeres, which of the retina, the three other dorsal ipsilateral blastomeres normally do not produce retina, were transplanted to the each produce about an eighth of the retina, and the four position of D 1.1.1. The progeny of ventral equatorial blas- contralateral dorsal blastomeres and an ipsilateral ventral- tomeres (Vl .1.2 and V2.1.2) populated the retina in a pattern animal cell together produce the remaining eighth of the that was indistinguishable from that of normal Dl. 1 .l, show- retina. There was no significant spatial segregation of the ing that position in the cleavage stage embryo is an important clones derived from different progenitors in either the an- determinant of retinal lineage. However, progeny of trans- terior-posterior or dorsal-ventral axes of the retina and no planted ventral- or dorsal-vegetal pole blastomeres (V2.1 .l boundaries between clones were observed. Instead, the and D2.1 .l) never populated the retina, demonstrating that clones intermixed to form multiple radial subclones that were not all blastomeres are competent to form retina. equivalent to those demonstrated by marking optic vesicle [Key words: Xenopus laevis, retina, determination, fegu- progenitor cells (Holt et al., 1988; Wetts and Fraser, 1988). lation, competence, transplantation] This mosaic pattern was initiated by the beginning of gas- trulation, advanced in the neural plate, and virtually complete The mechanismsby which a multipotent embryonic cell is de- in the optic vesicle. At optic vesicle stages cell movement termined to become one particular neuronal cell type is poorly within subclones was restricted, resulting in the formation understood in vertebrates, but it is generally believed that com- of lineally related columns of cells in the mature retina. mitment to a particular CNS phenotype occurs in a complex To determine if the blastomere progenitors are determined seriesof cellular decisionsthat affect the developmental poten- to produce these retinal lineage patterns, the major retinal tial of progenitor cells and the specificphenotypes of their prog- progenitor (D1.l.l) was deleted bilaterally. About 80% of eny (reviewed by Jacobson,199 1; McConnell, 1991). There are the tadpoles developed normal-appearing eyes; of these the severalexamples indicating that the developmental potential of retinas in two-thirds were normal in size and the rest were neuronal progenitors is gradually restricted. Cell lineagein ze- smaller. The blastomeres surrounding the deleted Dl .l .l brafish becomestissue specificduring gastrulation, but becomes progenitors changed their contributions to retina in different cell type specificlater (Kimmel and Warga, 1986).Cortical neu- ways to effect a complete or partial restoration. Ventral blas- rons becomecommitted to a laminar fate at the time of their tomeres, which normally contribute mainly to the tail, pro- final mitotic division, but are not committed to a particular cell duced substantial amounts of the retina while dorsal blas- type until later (McConnell, 1991). In the retina, progenitors tomeres, which normally contribute mainly to the head, may first be restricted to produce cells in certain regions(Wil- decreased their contribution to the retina. To determine liams and Goldowitz, 1992) and then, perhaps as late as the whether these changes in retinal lineage were due to changes terminal mitosis (Turner and Cepko, 1987; Holt et al., 1988) in blastomere position after the surgery, various other blas- to produce particular cell types. tomeres were deleted prior to lineage mapping. Dorsal-an- Since the retina is accessibleto lineage tracing techniques imal blastomeres took over the retinal fate of their dorsal- during development, and is composed of a variety of easily vegetal neighbors after those neighbors were deleted, but identifiable cell types that are arrangedin a highly ordered spa- did not change fate after the deletion of their ventral-animal tial pattern, it has served as a model for understanding the relationship betweencell lineageand cell type determination in Received Oct. 9, 1992; revised Feb. 16, 1993; accepted Mar.. 1, 1993. the nervous system. Labeling of mitotic precursor cells in the This work was supported by NIH Grants NS23 158 and EY 10096. We thank optic vesicle or perinatal retina, either by intracellular lineage Mrs. Lianhua Yang for her excellent technical assistance. dye injection (Holt et al., 1988; Wetts and Fraser, 1988; Wetts Correspondence should be addressed to Dr. Sally A. Moody, Department of et al., 1989)or by infection with a recombinant retrovirus (Price Anatomy and Neuroscience Program, The George Washington University Medical Center, 2300 I Street, N.W., Washington, D.C. 20037. et al., 1987; Turner and Cepko, 1987; Turner et al., 1990), Copyright 0 1993 Society for Neuroscience 0270-6474/93/133193-18$05.00/O elegantly demonstratesthat most of the clonesderived from a 3194 Huang and Moody - Lineage Regulation in Frog Retina single neuroepithelial cell contain more than one retinal cell To test experimentally whether these progenitors are deter- type; mitotic retinal cells rarely produce a clone consisting of a mined regarding these retinal lineage patterns, we bilaterally single phenotype. Although this fact has often been cited as deleted the major progenitor of the retina and quantified the demonstrating that cell lineage plays no role in vertebrate CNS retinal contribution of the remaining blastomeres. To test whether phenotype determination, this observation addresses only one blastomeres that normally do not produce retina are competent way that lineage may be important, that is, to produce a clone to do so if placed in the appropriate position, single labeled cells of cells having the same phenotype. There are several examples were either deleted from or transplanted into the site of the that show that cell lineage directs phenotype choices in other major retinal progenitor. The results from both approaches sup- ways (reviewed in Kenyon, 1985; Stent, 1985; Davidson, 1990). port the hypothesis that the position of a blastomere prior to Lineage patterns may direct different progenitors to produce gastrulation determines whether it will be a progenitor of the prescribed subsets of cell types, as has been shown for nematode retina, in most cases regardless of the blastomere’s original lin- motoneurons (White et al., 1982). Lineage genes may direct eage. However, some vegetal blastomeres are not competent to progenitors to divide a prescribed number of times and thus make retina even when located in the most “retinogenic” po- control the number of cells in a population (Chalfie et al., 198 1; sition. Williams and Herrup; 1985). Finally, lineage patterns may place the descendants of a particular progenitor in prescribed domains Materials and Methods of a tissue, such that their phenotypes are determined by region- Embryo collection. Fertilized Xenopus laevis eggs were obtained by go- specific cell-cell interactions. This occurs in the nematode vulva nadotropin-induced mating ofadult frogs. After the jelly coat was chem- ically removed, two- and early four-cell stage embryos with the first (Stemberg and Horvitz, 1989), and has been suggested in nu- cleavage furrow bisecting the gray crescent (Klein, 1987) were collected merous regions of the vertebrate CNS (Jacobson, 1983, 1985; for later use. At 32 cells, only those embryos with stereotypic radial Herrup, 1986; Fraser et al., 1990; Leber et al., 1990). cleavages (type X, Jacobson 198 1a; Moody, 198713) were used for lineage In order to investigate whether these other kinds of lineage dye injection, blastomere deletion, and transplantation. By selecting this patterns have a significant role in vertebrate CNS pattern for- population for study we minimized any variations in furrow pattern and thus kept blastomere position consistent among animals (Moody, mation, it is necessary to study the variation in phenotype, cell 1987a,b, 1989; Moody and Kline, 1990; Huang and Moody, 1992). In number, and spatial distribution between clones derived from other words, we were able to inject, delete, and transplant progenitors the same progenitor in many different animals. Analysis of clone that were nearly invariant among animals. variation has been a most important means for establishing the Lineage dye injection. About 1 nl of 0.5% Texas red-dextran amine (Molecular Probes) or fluorescein-dextran amine (Molecular Probes) role cell lineage plays in nematode development. In fact, the was pressure injected into identified blastomeres. Previous fate maps critical evidence that lineage has an important role in cell phe- of 32-cell embryos demonstrate that the cells in the retina are derived notype determination was the discovery that the mitotic history mainly from five animal pole blastomeres on each side of the animal of the same identified progenitor in many different animals is (Dale and Slack, 1987a; Moody, 1987b). According to Jacobson’s no- invariant (Sulston et al., 1983). However, studies using tech- menclature these five blastomeres are Dl. 1.1, Dl. 1.2, D1.2.1, D1.2.2, and V1.2.1 (Fig. IA; Jacobson and Hirose, 1981). According to Nak- niques that mark neuroepithelial cells randomly cannot label amura’s nomenclature these blastomeres are Al, B 1, A2, B2, and A3, the same progenitor repeatedly and therefore cannot demon- respectively (Nakamura et al., 1978). For our quantitative maps, sur- strate if the clone of a given progenitor is invariant in terms of rounding neighbors also were injected in order that no potential or minor cell phenotype, cell number, or regional distribution. progenitor be missed. In this report we took advantage of the existence of stereo- Tissue processing. The embryos were fixed by immersion in 0.1 M phosphate-buffered saline (PBS) containing 4% naraformaldehvde and typed progenitor cells at the cleavage stages of frog development 3% sucrose at stages lo-25 anA 43-44 (Nyeuwioop and Fabe;, 1964) that can be repeatedly labeled in many different embryos. We and were cut at 14 pm in transverse or sagittal serial sections with a found that in the normal embryo the retina descends from a cryostat. Sections were collected on slides, washed in 0.1 M PBS, and restricted and invariant subset of those blastomeres that produce coverslipped in 95% glycerol. Quantitative analysis of clones in the retina. The data for the quan- the forebrain. By quantitatively analyzing many clones derived titative analyses were collected from the retina at stages 43-44, when from the same progenitor, we show that each retinal progenitor the neurogenesis of the retina is almost completed and the retinal layers produced about the same proportion of the different retinal and the cell types are easily identified (Holt et al., 1988). The retina was phenotypes. Each progenitor produced a distinct amount of ret- divided into anterior, middle, or posterior thirds, and dorsal, central, or ventral thirds with reference to the position of the lens. The number ina, but the precise numbers of cells in the clones varied greatly of fluorescently labeled cells in each third was counted from every other in different embryos, demonstrating that the number of mitoses tissue section, and accordingly the total clonal cell numbers and distri- is not restricted within a lineage. We found no evidence for bution to retinal layers were calculated. In addition, these sections were clonal or laminar boundaries between clones, demonstrating photographed and the area of the retina on the prints was measured that blastomere origin does not restrict the spatial distribution with the Bioquant System IV image analysis program. Based on these measurements the total volume of the retina and the cell density of of clones within the retina. All blastomere clones formed mul- labeled clones derived from each blastomere were calculated. Only em- tiple discrete cell columns, which extended radially across retinal bryos in which individual cells were discemable were included in the layers, that were equivalent to later-marked clones (Holt et al., cell counts; at’least five embryos were counted for each blastomere per 1988; Wetts and Fraser, 1988; Turner et al., 1990). We show manipulation. that the columns descended from one blastomere were inter- Blastomere ablation and transplantation. The detailed procedures of blastomere deletion and transplantation have been described in previous mixed extensively with columns descended from the other blas- reports (Gallagher et al., 199 1; Huang and Moody, 1992). In this part tomeres. This intermixing began during gastrulation and was of the study three groups of lineage dye injection and experimental virtually complete by the optic vesicle stages, at which time operations were performed to test blastomere determination and com- retinal cells begin to become postmitotic. At this latter stage, petence regarding retinal cell lineage. First, to test whether a normal retina can develop when an embryo does not contain its full complement cell movement within subclones was restricted, resulting in the of retinal prdgenitors, both D 1.1.1 blastomeres were removed from the formation of lineally related columns of cells in the mature embryo with sharpened forceps (Fig. 1C,D). Care was taken so that no retina. cellular debris remained in the wound, which could affect the regulative The Journal of Neuroscience, August 1993, 13(E) 3195

1A

'1 4 F %--

Figure I. A-D, Diagrams and photographs of a 32-cell embryo (animal pole view) before (A C) and immediately after (B, D) D 1.1.1 deletion. The nomenclature of the blastomeres is after Jacobson and Hirose (198 1). The thick arrows point to the midsagittal cleavage. The thin arrows in B indicate the relative changes in retinal lineage for each blastomere after deletion of both Dl . 1.1 (indicated by dotted patch in B and an asterisk in D). The dash in some blastomeres in B indicates no change in retinal lineage (see also Fig. 5). Scale bars, 500 pm. E, Two examples of stage 43 embryos with normal-appearing eyes in which both D1.l. 1 were deleted. Scale bar, 1000 pm. F, Lateral view of a 32-cell embryo showing the blastomere tiers and all the vegetal cells used in the transplantation studies. A, animal pole; Vg, vegetal pole. The dorsal side is to the top and the ventral side is to the bottom. 3196 Huang and Moody - Lineage Regulation in Frog Retina

Table 1. The number of retinal cells that descend from each blastomere of 32-cell embryos

Cell Cell number density % Blastomere (mean + SEM) O/aTotal Range (cells/mm2) Total D1.l.l ips? 4714 iz 600 49.1 3620-7054 7194 t 2239 43.1 D. 1.1.1 contrab 668 + 348 7.0 196-2028 1411 f 741 8.6 D1.1.2 ipsi 1338 k 402 13.9 258-2748 2006 + 1102 12.2 D1.1.2 contra 442 I+ 176 4.6 lo-1034 1096 k 380 6.7 D1.2.1 ipsi 1136 t 362 11.8 242-1957 2336 f 1099 14.2 D1.2.1 contra 36 t- 30 0.4 o-155 137 f 119 0.8 D1.2.2 ipsi 1116 -t 432 11.6 384-2699 2140 f 1328 13.0 D1.2.2 contra 34 i 20 0.4 O-92 61 + 39 0.4 V1.2.1 ipsi 62 & 8 0.6 42-90 71k48 0.4 V1.2.1 contra 0 y Ipsilateral. hContralateral. abilities of the remaining cells (Roux, 1888). To demonstrate which of from ipsilateral V 1.2.1, contralateral D 1.2.1, and contralateral the remaining blastomeres regulate their lineages, one or two blasto- D 1.2.2. These differences also are observed if the cell numbers meres in one embryo were injected with 1 nl of lineage dye, and then from the different blastomeres are converted to retinal cell den- both Dl. 1.1 blastomeres were removed. As a control, only the vitelline membrane was removed from embryos after lineage dye injection. sity (cells/mm$ Table l), which eliminates any variation due In a second set of experiments, several combinations of dye injection to differences in the number of cells generated at the different and cell deletions were performed to determine whether positional embryonic stages (43 vs 44). changes of the remaining blastomeres account for the changes in retinal As described previously for other cell types (Moody, 1987a,b, lineages observed in the first experiment. (1) Deletion of both Vl. 1.1 blastomeres was combined with Dl. 1.1 injection to see if Dl. 1.1 de- 1989; Huang and Moody, 1992), each blastomere usually pro- scendants move ventrally to adopt the V 1.1.1 fate and contribute less duces a consistent clonal pattern among embryos, but this pat- to the retina. (2) Unilateral deletion of D 1.1.2 was combined with D 1.1.1 tern is not invariant. That is, the number and location of the injection to see if Dl . 1.1 descendants move dorsal-vegetally to adopt labeled cells may vary from embryo to embryo after injection the D 1.1.2 fate and contribute less to the retina. (3) Unilateral deletion of the same blastomere progenitor. For example, after D 1.1.1 of D2.1.2 was combined with Dl. 1.2 injection to see if the descendants move into the vegetal hemisphere to adopt the D2.1.2 fate and no longer injection labeled cells always were in both retinas, but the size produce retinal cells. of the ipsilateral clone in one embryo could be twice as large as To test directly whether blastomere position determines retinal fate, that in another embryo (Table 1). Furthermore, the number of the midline cells that normally do not contribute to the rostra1 CNS or cells in each clone from a particular blastomere varied from one retina (V 1.1.2, V2.1.2, V2.1.1, D2.1.2, and D2.1.1; Fig. 1F) were labeled with lineage dye and then transplanted into the position of Dl.l.l in embryo to another. The range in cell number was only about an unlabeled host. Single donor blastomeres were transplanted into the twofold for ipsilateral D1.l. 1 and V1.1.2, but was about lo- space of bilaterally removed D 1.1.1 to accommodate the bigger donor fold for the other blastomeres (Table 1). Therefore, retinal pro- cells. In addition, labeled D 1.1.1 was transplanted into the V 1.1.2 po- genitors are not determined to produce a set number of cells; sition to test whether it continues to produce retina in a novel position. As a control, labeled D 1.1.1 was orthotopically transplanted. Only em- that is, the different retinal lineages are not each characterized bryos with normal retinal structure were collected for the quantitative by an invariant number of mitoses. analyses. There are no clonal boundaries in the retina Results In order to discern whether ancestry restricts the spatial distri- Blastomere progenitors contribute d$erent amounts of the bution of clones, the cell density of different clones in six retinal retina thirds (anterior, middle, posterior, dorsal, central, ventral) was The present study confirms the findings of previous fate maps measured. The descendants from each blastomere were neither (Dale and Slack, 1987a; Moody, 1987b) that five animal hemi- evenly distributed throughout the retina, nor restricted to a sphere blastomeres (D1.l.l, D1.1.2, D1.2.1, D1.2.2, V1.2.1; particular sector of the retina (Figs. 3-5). Rather, they often Fig. lA, Table 1) on each side of the 32-cell embryo give rise were distributed.in a shallow gradient across the retina. In the to the retina. This is a restricted subset of those blastomeres ipsilateral retina, Dl.2.1, D1.2.2, and V1.2.1 contributed more that normally contribute to the forebrain (Huang and Moody, to the anterior (nasal) third, while D 1.1.1 and D 1.1.2 contrib- 1992). By precisely quantifying each blastomere’s contribution uted slightly more to the posterior (temporal) third of the retina to the retina, we show that no other blastomere has any progeny (Fig. 3A). The clone derived from D1.2.1 showed the largest in the retina. Each of the five retina-producing blastomeres con- regional gradient: this blastomere contributed 10 times more to tributes a much larger number of ipsilateral than contralateral anterior than to posterior retina (p < 0.05, t test). In the con- cells, and each contributes different numbers of cells to the retina tralateral retina, no anterior-to-posterior gradients were dis- (Fig. 2A, Table 1). On the average, each retina descends 50% cerned. In the ipsilateral retina, most of the clones were evenly from ipsilateral D 1.1.1, 14% from ipsilateral D 1.1.2, 12% from distributed in the dorsal-ventral direction, although D 1.2.2 had ipsilateral D1.2.1, 12% from ipsilateral D 1.2.2, 7% from con- slightly fewer cells on the ventral side (Fig. 4A). In the contra- tralateral D 1.1.1, 5% from contralateral D 1.1.2, and < 1% each lateral retina the clones derived from D 1.1.1 and D 1.1.2 were The Journal of Neuroscience, August 1993, i3(8) 3191

Table 2. Percentage of descendants derived from each blastomere in D112c D121c D122c each retinal layer es...- 4.6% \ 0.4% 1 / 13.4%

Total cells N PR INL CC counted D1.l.l 5 24.6 50.7 24.7 10,334 D1.1.2 5 19.7 55.4 24.9 3568 D1.2.1 5 27.9 48.4 23.7 1509 D1.2.2 5 20.5 57.8 21.7 2421 v1.2.1 5 23.1 50.0 26.9 52 Overall 25 23.2 52.4 24.4 17,820 Whole retina0 24.0 54.0 22.0 The data were gathered from the middle third of the retina where the cross section of the tissue illustrates the retinal layers most clearly. The data are expressed as the means from five embryos per blastomere. With a x2 analysis, there is no difference in distribution between the blastomeres (p > 0.25). a Data from Holt et al. (1988). distributed more in the ventral third. Although there were some shallow gradients of clone distribution along the retinal axes, descendantsof particular blastomereswere not restricted to par- ticular regions of the retina; all regions of the retina were of polyclonal descent(Fig. 5). There is no preferred clonal laminar distribution Previous studiesin Xenopus demonstrated that the majority of clones derived from optic vesicle progenitor cells span two or D121 three retinal layers (Holt et al., 1988; Wetts and Fraser, 1988; 1.9% Wetts et al., 1989), and thus are not restricted by lineageto a particular cellular lamina. However, since mammalian visual cortical cells are restricted to whichever lamina they will pop- ulate before their phenotype is restricted (McConnell, 199l), it seemedpossible that the different retinal lineagescould have preferred, if not restricted, laminar destinations. By counting the number of cells in each retinal cell layer derived from each blastomere in a large number of animals, we observed no dif- ference among the blastomeres(Table 2). For all blastomeres, inner nuclear layer (INL) cells account for approximately half of the descendantsof eachblastomere and photoreceptors (PR) and ganglion cells (GC) each account for one-fourth. These fig- d112i ures are closeto the overall proportion of the different cells in 0.3% the whole Xenopus retina (Holt et al., 1988). This result dem- Figure2. A, A pie graph showing the ipsilateral (i) and contralateral onstrates that each blastomere clone is representative of the (c) contributions to the normal retina from each blastomere. Ipsilateral entire retina, in terms of cellular phenotypes, and that none of D 1.1.1 is the major contributor, producing about half of the retina; the the blastomeresplace descendantsin a preferred lamina. How- other three dorsal blastomeres each produce about one-eighth of the retina, and the remaining one-eighth descends from contralateral dorsal ever, as discussedbelow, the laminar distribution of cells within blastomeres and one ventral blastomere (V1.2.Ii). B, A pie graph show- some radial subclonesfrom different blastomeresis different ing the clonal composition of the retina in embryos from which both from that of the whole retina. D 1.1.1 cells were deleted. Together the V 1.2.1 blastomeres become the major contributors, producing more than half of the retina; this ap- The retina is a mosaicof radial subclones proximates the normal contribution of both D 1.1.1 (compare to A). Blastomeres D1.2.1 and V 1.1.1 produce most of the rest of the retina. The descendantsof each blastomere in the stage43-44 retina did not form a coherent massseparated from the descendants of other blastomeresby clonal boundaries. Instead, they were a solid cell mass,as is the casefor clonal cohorts in the retina dispersedin many discrete clusters that were intermixed with of chimeric mice (Williams and Goldowitz, 1992) and mice similar clustersdescended from other blastomeres(Fig. 5). Each marked with retrovirus at early embryonic days (Turner et al., cluster was composedof radial columns that spannedall retinal 1990). Instead, the Xenopus cell cluster was dominated by la- layers and were very similar in appearanceto the larger clones beledcells, but also contained unlabeledcells derived from other initiated in the optic vesicleby either retroviral infection (Turner blastomeresin any of the retinal layers (Fig. 5; see also Wetts and Cepko, 1987; Turner et al., 1990) or intracellular injection et al., 1989). In addition, a few labeledcells lay betweenlabeled (Holt et al., 1988; Wetts and Fraser, 1988; Wetts et al., 1989). clusters (Fig. 5D), and their cluster membership could not be Each cell cluster, although having discrete boundaries, was not ascertained.As proposedin the mouse(Williams and Goldow- 3198 Huang and Moody * Lineage Regulation in Frog Retina

N - 10000 8000

z 3 6000 6000 iii 0, 6000 E I- 5 B 4000

I- i ti 2000

1’ Dill D112 D121 D122 v121 D112 Dl21 0122 Vlll v121 v122

10000 ANTERIOR MIDDLE 1 q POSTERIOR

6000

i 2000 5

A ’ Dill D112 1.1121 D122 ~121 B ’ D112 D121 D122 ~111 ~121 ~122

Figure 3. A, The blastomeres’ regional contributions to anterior, middle, and posterior thirds of the normal retina are presented in terms of cell density. Ipsilaterally, D 1.1.1 and D 1.1.2 contribute more posteriorly, while D1.2.1, D 1.2.2, and V1.2.1 contribute more anteriorly. However, the only significant regional difference in cell density was observed in the D 1.2.1 clone (*, p < 0.05, t test). No contralateral differences were observed. B, The regional contribution of each remaining blastomere to anterior, middle, and posterior thirds of the retina in Dl.l.l-deleted embryos. Ipsilaterally, V1.l. 1 and V1.2.1 contribute slightly more to the anterior retina, and contralaterally, V1.2.1 contributes slightly more to posterior retina. No regional differences were observed for the other blastomeres. itz, 1992) we suggestthat these clusters represent radial units V1.2.1 subclonedid not contain PRs.Thus, the radial subclones or subclonesthat descendfrom separate neuroepithelial pre- in our material most likely represent the progeny of a single cursors. neuroepithelial cell that producesall retinal cell types. It should When a blastomere produced many subclonesin the retina, be noted that these subclonesare all different in size, demon- the boundariesbetween subcloneswere not distinct (e.g.,as seen strating (asin Table 1) that the number of mitosesin subclones after D 1.1.1 injection; Fig. SO), but in some casesthere were is not invariant. only one or a few subclonesin the entire retina (e.g., Fig. SA- Interestingly, although the proportion of cells in the different C). In thesecases the cell number and cell type composition of laminae in the entire retinal clone of every blastomerewas quite the radial subcloneswere quantified (Table 3). In contrast to similar to that of the whole retina (seeprevious section and radial cells marked late in development (Holt et al., 1988;Wetts Table 2) there was considerable variation from this pattern and Fraser, 1988),almost every subclonecontained every retinal within individual radial subclones(Table 3). For example, the cell type. Rather than identifying each cell in a clone by mor- three Dl. 1.2 subclonescontained a large proportion (82%) of phology, we categorized retinal cells according to the layers in cells in the INL and a very small proportion (2%) in the GC which they resided, using the criteria of Holt et al. (1988). In layer. The D 1.2.2 and V 1.2.1 subclonescontained slightly more the INL, cells were subdivided as belonging to one of three cells in the INL and slightly fewer PRs than the whole retina. sublayers:cells on the vitreal side (I,) are mainly amacrinewith The boundaries between subclonesfrom Dl. 1.1 rarely were a few interplexiform cells, cells in the middle (I,) are mainly sufficiently distinct to analyze, and thus only one is presented bipolar and Miiller cells, and cells on the epithelial side (I,) are (Table 3); that particular clone contained the highestproportion mainly horizontal and bipolar cells. In 13 out of 16 casesthe of PRs.These results indicate that severalof the radial subclones radial subclonecontained cells in all five retinal sublayers.As show preferred laminar distributions, which loosely correlated exceptions, two Dl . 1.2 subclonesdid not contain GCs and one with blastomereof origin. The Journal of Neuroscience, August 1993, 73(8) 3199

8000-

6000 -

Ill11 D112 Dl21 0122 VI21 0112 Dl21 D122 Vlll v121 v122

10000 DORSAL DORSAL CENTRAL CENTRAL k?j VENTRAL •j VENTRAL 8000 I

B ' D112 D121 D122 ~111 ~121 ~122 Figure 4. A, The blastomeres’ regional contributions to dorsal, central, and ventral thirds of the normal retina are presented in terms of cell density. No obvious distribution gradients were observed on the ipsilateral side, but on the contralateral side dorsal blastomeres contribute more to the ventral retina. B, The regional contribution of each remaining blastomere to dorsal, central, and ventral thirds of the retina in Dl . 1.1-deleted embryos. D 1.2.1 and V 1.1.1 contribute slightly more to the ipsilateral dorsal retina. V 1.2.1 contributes slightly more to central retina, both ipsilateral and contralateral.

clonesfrom Xenopusoptic vesicle cells (Holt et al., 1988; Wetts Early mixing of blastomere clones and Fraser, 1988), the blastomereclones already have formed Since the subclonesfrom the blastomere progenitors are inter- a lineagemosaic. The radial subclonesin the optic vesicle con- mixed as radial clustersin the stage43-44 retina, and sincethe sistedof either (1) one single,columnar, neuroepithelial cell that clones initiated from the optic vesicle neuroepithelial cells are spansthe vesicle; (2) small groupsof two or three such cells; or restricted to radial clusters(Holt et al., 1988; Wetts and Fraser, (3) small clustersof cells composedof columnar neuroepithelial 1988) a restriction in cell mixing seemsto occur prior to the cells and cells that have withdrawn a processfrom either the final mitosis. We examined the pattern of the clones derived epithelial or vitreal surface(Fig. 6). Theselatter cells appeared from the different blastomere progenitors during optic vesicle to be postmitotic and were observed only in the larger clusters. formation, neurulation, and gastrulation in order to determine It is significant that at the time when retinal cells begin to leave when clonal mixing starts, when it ceases,and when the radial the mitotic cycle, subclonesderived from neuroepithelial pre- subclonesare established. cursors remained coherent; there is little lateral migration of The basic pattern of blastomereclones in the stage25 optic cells within the mitosing population. vesicle, at which time retinal cells begin to be born (Holt et al., Examination of earlier stagesalso revealed considerablemix- 1988), closely resembledthat of the stage43-44 retina, at which ing between different lineagesin the prospective retinal areas. time nearly the entire complement of embryonic retinal cells In the neural plate (stage 14) the retinal field is located at the are postmitotic (Holt et al., 1988). That is, the labeled cells most rostra1part of the plate and includes the adjacent neural derived from one blastomere were dispersedthroughout the crest (Brun, 1981; Eaglesonand Harris, 1990). The labeled co- entire optic vesicle and were intermixed with cells derived from lumnar cells in the retinal field were well separatedfrom one other blastomeres(Fig. 6). The density of the radial subclones another by the progeny of other blastomeres(Fig. 7A). The was similar to that observed in the stage43-44 retina (compare proportion of cells from the different blastomereswas similar Figs. 5, 6). Thus, at the time when previous studies initiated to their final contribution to the differentiated retina. For ex- bang and Moody l Lineage Regulation in Frog Retina

Figure 5. Photomicrographs of radial subclones in the stage 43-44 retina. ph, photoreceptor layer; id, inner nuclear layer; g, ganglion layer. A, One radial subclone derived from contralateral Dl . 1.1 contains cells in all retinal layers. Within the cell cluster, there are some unlabeled profiles, indicating that cell mixing between lineages has occurred. B, Two small radial subclones derived from ipsilateral D 1.2.2 each contain cell types in all retinal layers. These subclones form coherent columns. C, Three radial subclones (brackets) derived from ipsilateral D1.2.1 are dispersed far apart in different retinal sectors. All retinal cell types are contained in the subclone in the dorsal sector (rap). D, After ipsilateral Dl . 1.1 injection many labeled descendants are distributed throughout the retina. The segregation of cell clusters into radial subclones is not as clear as in other cases (compare A-C) and a few cells (arrows) lay between the labeled clusters. However, five columnar groups can be recognized (brackets). Scale bar, 50 pm. The Journal of Neuroscience, August 1993, 73(8) 3201

Table 3. Percentage of cells in each retinal layer derived from different radial subclones

Total Total Blastomere number PRa ISh kc IId I GC D1.l.l contrd 136 36.0 19.1 13.2 13.2 (45.6) 18.4 D1.1.2 ipsig 42 9.5 11.9 28.6 43.9 (84.4) 7.1 D1.1.2 ipsi 8 12.5 50.0 0 37.5 (87.5) 0 D1.1.2 contra 8 25.0 12.5 37.5 25.0 (75.0) 0 D1.2.1 contra 71 28.2 5.6 23.9 26.8 (56.3) 15.5 D1.2.1 contra 148 23.6 18.9 18.9 22.4 (60.2) 16.2 D1.2.1 contra 32 28.1 9.4 34.4 6.3 (50.1) 21.8 D1.2.2 contra 36 19.5 13.9 33.3 25.0 (72.2) 8.3 D1.2.2 contra 30 16.7 10.0 23.3 26.6 (59.9) 23.4 D1.2.2 ipsi 17 41.2 11.8 23.5 17.6 (52.9) 5.9 D1.2.2 ipsi 10 10.0 10.0 20.0 30.0 (60.0) 30.0 V1.2.1 ipsi 47 27.6 8.5 25.6 25.5 (59.6) 12.8 V1.2.1 ipsi 26 23.1 19.2 23.1 26.9 (69.2) 7.7 V1.2.1 ipsi 12 16.7 25.0 16.7 33.3 (75.0) 8.3 V1.2.1 ipsi 11 0 18.2 36.3 27.3 (81.8) 18.2 V1.2.1 ipsi 104 31.7 14.5 19.2 15.4 (49.1) 19.2

Mean 26.3 15.0 21.4 21.7 15.6 n Photoreceptors. h Inner nuclear layer 3, consisting of mainly horizontal and bipolar cells. c Inner nuclear layer 2, consisting of mainly bipolar and Mtiller cells. d Inner nuclear layer 1, consisting of mainly amacrine and interplexiform cells. c Ganglion cells. i Contra = clone from contralateral side. d Ipsi = clone from ipsilateral side. ample, many labeled cells in the retinal field were from Dl . 1.1 In summary, although in the majority of casesthe remaining and very few were from V1.2.1 (Fig. 7B). In the gastrula the blastomeresregulate to produce a normal appearingeye, often retinal area lies in the prospective anterior neural area (Keller, they do not completely compensatefor the removal of the major 1975). The columnar cells derived from the different retinal retinal progenitors. blastomereprogenitors in this region as early as stage10 formed a mosaic with the descendantsfrom other blastomeres(Fig. 8). Remaining blastomeresrespond d@erently to the bilateral However, the mixing was less extensive than at neural plate deletion of 01.1.1 stages,suggesting that mixing begins only shortly before the To define which of the remaining blastomeresregulated to re- onset of gastrulation. Theseresults indicate that the prospective store the retinal lineages,we injected the different blastomeres retinal field becomes a lineage mosaic well before neural in- with lineage dyes before the removal of both Dl. 1.1 and then duction takes place. quantified their clonal contribution to the retina. None of the tier 3 blastomeres(Fig. 1F) nor the tier 2 ventral midline blas- The retina can be restored after its major progenitor is deleted tomere (V 1.1.2) gave rise to retina (Fig. 1B). However, almost D 1.1.1 produces over 50% of the cells in the retina. After de- all other animal blastomereschanged fate, some by increasing letion of both D 1.1.1, 60% of the embryos developed normally their contribution to retina and some by decreasingtheir con- (Table 4) and had normal-appearing eyes (e.g., Fig. 1E). The tribution to retina (Fig. 1B). Ipsilateral D1.2.1 (ipsilateral with retinal layers were well organized and all of the retinal cell types respect to the retina analyzed) remained one of the major con- could be identified (Fig. 9). However, theserestored retinas were tributors, but most of the retina now descendedfrom ventral not all the same size as those of normal embryos (Fig. 10). Of cells (Fig. 2B). V1.2.1 became the major retinal progenitor, the embryos that were not externally normal (40%), half had producing 40% of ipsilateral and 15% of contralateral retina. eye defects (Table 4), which included no eyes (0.5%) only one This approximatesthe normal contribution of the deletedD 1.1.1. eye (7.0%), much smaller eyes(7.0%) or histologically distorted Ipsilateral V 1.1.1, which normally doesnot contribute to retina, eyes (4.7%), and half had body defects, which included an open produced retinal cells in 12 of 13 casesand becameone of the dorsal axis (7.0%) or failure to gastrulate(13.9%). In the control largest contributors. As compared to normal cell counts, ipsi- group.,in which only the vitelline membranewas removed, 4.7% lateral V 1.1.1 and V1.2.1 and contralateral V1.2.1 significantly of the embryoshad eyedefects and 9.3%had body defects(Table increasedtheir contributions to retina, whereasipsilateral D 1.1.2 4), indicating that the operational procedurecaused nonspecific and D1.2.2 significantly decreasedtheir contributions (Fig. 11). damage. However, eye defects occurred in the deletion cases BlastomereV1.2.2, which never contributes to retina in normal 15% more frequently than in the controls, indicating that the embryos, gave rise to a small number of retinal cells in two of removal of the two major retinal progenitors can prohibit full 13 cases.As shown for the normal retina (Table l), the contri- retinal development. bution to the retina of the same blastomere after deletion of

The Journal of Neuroscience, August 1993, 13(E) 3203

Table 4. Summary of the development of embryos after bilateral Table 5. The number of retinal cells that descend from each Dl.l.1 deletion blastomere after deletion of both D1.l.l

Percentage of embryos Blastomere Mean + SEM Range % Total Eye Body (n) Number Normal defects defects D1.1.2 (5) Ipsi 16 + 16 O-84 0.3 D 1.1.1 -deleted 172 59.9 19.2 20.9 Contra 0 0 0 Control 43 86.0 4.7 9.3 Dl.2.1 (6) Ipsi 1206 * 360 322-2632 24.3 Contra 96 zk 96 O-564 1.9 The eyedefects category includes embryos with no eyes,only one eye,microphthal- mia, or histologically distorted eyes.The body defectscategory includes embryos D1.2.2 (6) Ipsi 94 k 62 O-378 1.9 with an open dorsal axis or embryos that failed to gastmlate. The control group Contra 0 0 0 had the vitelline membrane removed only. Vl.l.l (5) Ipsi 782 + 362 114-1894 15.7 Contra 24 + 22 O-116 0.5 D 1.1.1 varied a lot from individual to individual (Table 5) V1.2.1 (5) Ipsi 2144 + 548 1062-3722 40,.5 confirming that the blastomereprogenitors are not restricted to Contra 732 k 314 o-1 544 14.7 produce fixed numbers of retinal cells. V1.2.2 (5) Ipsi 10 * 10 O-50 0.2 The descendantsof the new blastomere progenitors formed Contra 0 0 0 radially oriented columns across all retinal layers and all the subclonescontained more than one type of retinal cell (Fig. 9) as described for the normal retina. To determine whether de- scendantsfrom the new progenitorswere distributed in different increasedtheir contribution to the retina, while those dorsal- regions of the retina, the cloneswere quantified in anterior-to- vegetal to the deleted cells decreasedtheir contribution (Fig. posterior and dorsal-to-ventral retinal thirds. Vl. 1.1 and V1.2.1 1B). Thesetopographically related lineagechanges suggest that contributed slightly more to anterior than to posterior retina, the deletion of D 1.1.1 may have caused a general dorsal-to- which was complemented by contralateral V1.2.1 contributing vegetal shift in the position of the remaining blastomeres,which slightly more to posterior retina (Fig. 3B). V 1.1.1 and D1.2.1 in turn causedblastomeres to manifest a retinal fate according contributed slightly more to the dorsal third and V 1.2.1 slightly to their new positions. To test this possibility, we deleted dif- more to the central third of the ipsilateral retina (Fig. 4B). How- ferent animal blastomeresto shift the normal position of the ever, none of thesedifferences were statistically significant. The blastomeresin a predictable direction. Three different deletions regional distribution gradient of someblastomeres (e.g., D1.2.1) were done. changedin the D 1.1.1-deleted embryos, demonstrating that the Deletion ofbilateral Vl.1.1 combined with 01.1.1 injection (n spatial distribution of a clone is not yet fixed. As in normal = 14). This operation wasdesigned to seeif the dorsally located embryos, although there are shallow regional gradients of some D 1.1.1 shifts its position ventrally to adopt the fate of VI. 1.1, of the clones, descendantsfrom all blastomerescould be found which normally does not give rise to retina. After V1.l. 1 de- in any region of the retina, and no boundarieswere seenbetween letion, half of the embryos developed normal headsand eyes; clones (Fig. 9). in these embryos D 1.1. l’s contribution to retina was quanti- tatively comparable to that in normal embryos (Fig. 12A). The changes in retinal lineages of the remaining blastomeres Therefore, D 1.1.1 does not take the retinal fate of its ventral are position dependent neighbor. The regulatory responseof the remaining blastomeresto dele- Deletion of unilateral 01.1.2 combined with 01.1.1 injection tion of the D 1.1.1 cells differed according to the position of the (n = 12). This operation was designedto see if D 1.1.1 could blastomere in the embryo. Those ventral to the deleted cells take a more dorsal-vegetal retinal fate given the opportunity to c Figure 6. Thesephotomicrographs show column-like cell clustersin coronalsections of stage25 optic vesicle(ov). fl, forebrain;v, ventricle.A, Two Texas red-dextran amine-labeled cells (red) derived from ipsilateral V1.2.1 are well separated. One of them (large arrow) spans the entire retina and resembles a progenitor cell in the optic vesicle observed directly after injection (cf. Holt et al., 1988; Wetts and Fraser, 1988). The other cell (small arrow) has been split between two tissue sections. These single cells probably produce the well-separated radial subclones seen in stage 43-44 retina (compare 5, A or B). B, Ipsilateral Dl. 1.1 produces many descendants in the stage 25 optic vesicle (ov). Labeled and unlabeled subclones are intermixed, forming an ordered mosaic pattern. The labeled subclones form discrete cell columns, which are composed of only one cell (small solid arrow), a group of radial columnar cells (large solid arrow; the nuclei of individual cells are pointed out with arrowheads), or a group of cells with varying morphologies, including rounded without end feet on a limiting membrane (large open arrow). This micrograph illustrates the different kinds of progenitors available when lineages are initiated at the optic vesicle stage (e.g., Holt et al., 1988; Wetts and Fraser, 1988). Cells at the superior and inferior rims of the retina that appear to have neither vitreal nor pial contacts do not seem to fit the above description due to plane of section artifact. Scale bar, 50 pm. Figure 7. Photomicrographs of the prospective retinal area (brackets) of the neural plate (stage 14). e, ectodenn; m, mesoderm. Both are parasagittal sections with rostra1 to the right. A, The green cells descended from D 1.1.1 are intermixed with unlabeled cells derived from other blastomeres in the prospective retinal area. B, The cells derived from D 1.1.1 (green cells) and those few derived from V 1.2.1 (red cells) are intermixed with unlabeled cells from other blastomeres in the prospective retinal area. Scale bar, 50 pm. Figure 8. Photomicrographs of parasagittal sections through the anterior neural area of gastrulae. The right edges of the pictures are the halfway point between the animal and vegetal poles, and the animal pole is to the left. e, ectoderm; m, mesoderm. A, The red cells located at the edge of the D 1.1.1 clone are intermixed with many unlabeled cells from other progenitors at stage 11. The labeled cells are mostly arranged as columnar cells in the outerectodermal layer (e), resemblingthe radialcolumns in the future retina.B, At stage10 the green D 1.2.1descendants already are well mixed with red D1.2.2 descendants in the ectoderm (e) and the underlying mesoderm (m) in the anterior neural area. The border between the ectoderm andmesoderxn is indicatedby a broken line. Scalebar, 50 pm. 3204 Huang and Moody - Lineage Regulation in Frog Retina

Figure 9. Photomicrographs of stage 43 retinas from D 1.1.1 -deleted embryos showing that there are normal retinal layers and clonal patterns. A, Most of the retina is composed of radial subclones derived from V 1.2.1, which in normal embryos contributes very little to the retina. L, lens. B, Descendants of V 1.1.1, which in normal embryos produces no retinal cells, form a radial subclone that spans all retinal layers. Most retinal cell types can be identified in this subclone: ph, photoreceptor; b, bipolar cell; a, amacrine cell; g, ganglion cell. Scale bar, 100 pm. shift its position in that direction. Ten embryos in this group directly visualized; to test directly whether changesin position developed normally; two had smaller eyes. D 1.1. l’s contribu- affect retinal fate, different single-celltransplantations were per- tion to retina decreasedsignificantly in the 10 normal-appearing formed. embryos (Fig. 12B). The small retinal contribution of D 1.1.1 is Severalblastomeres produced retina when placedin the D 1.1.1 quantitatively similar to that of normal D 1.1.2 (Fig. 12C). This position. The deletion experiments described above demon- similarity suggeststhat after the dorsal-vegetal neighbor is de- strate that V 1.1.1 will produce retina when the D 1.1.1 position leted, D 1.1.1 shifts toward the vegetal pole and, in turn, adopts becomesavailable. In addition, Vl. 1.2, which does not con- the retinal fate of Dl . 1.2. We have not done the lineagetracing tribute to retina in normal embryos or in the Dl. 1.l-deleted experimentsto showwhich blastomerescompensate for Dl . 1.l’s embryos (Fig. 1B), producedthe D 1.1.1 pattern of labeledclones reduced contribution to retina. in all casesin which it was transplanted to the D 1.1.1 position Deletion of unilateral 02.1.2 combined with 01.1.2 injection (Table 6). The clones were found in the bilateral ventral fore- (n = 15). This operation was designedto seeif another animal brain and both retinas, with the majority of the labeled cells in blastomere could adopt a more posterior fate when given the the ipsilateral side (Fig. 13). The number of labeledcells in the opportunity to shift into the vegetal hemisphere.In this group retinal clone wascomparable to that of normal D 1.1.1 (Fig. 14), 13 embryos developed normally; in these casesD1.1.2’~ con- and the size of the retina in the transplant embryos was the tribution to retina was profoundly reduced (Fig. 120). Sixty sameas in normal embryos (p > 0.1; data not shown).A similar percent had no labeled cells in the retina, and the remainder result was obtained when V2.1.2, the ventral tier 3 blastomere had only a few labeled cells each. This retinal fate is comparable (Fig. lfl, was transplanted to the D1.l.l position (Table 6). to the normal fate of D2.1.2 to not contribute to retina. In Finally, even D 1.1.1 must be in its normal dorsal midline po- contrast, Dl .1.2 doesnot adopt the fate of its animal neighbor sition in order to produce retina. In a previous study (Gallagher (Dl. 1.1) when given the opportunity to shift its position in that et al., 1991) we transplanted D1.l.l to the tier 3 and tier 4 direction (Figs. 2, 11). ventral midline; in thesecases, although it produced spinal cord This set of experiments suggeststhat when cells are removed and axial mesoderm as is its normal fate, it never produced from the dorsal midline, there is a dorsal-to-vegetal rearrange- retina. Likewise, in this study we placed Dl. 1.1 in the position ment in the position of the remaining cells. Second, the re- of V 1.1.2 and no labeled cells were found in the retina or fore- maining cells take on the retinal fate of their new position. brain. Thus, all theseblastomeres can produce retinal cells when However, the positional changesof the blastomereswere not they are located in the correct position (dorsal-animal midline). The Journal of Neuroscience, August 1993, 13(8) 3205

RETINAL VOLUMES IN THE NORMAL AND l-l- THE D1.l.l ABLATED EMBRYOS UPSULA TERAL * * pco.01 * peo.1

j 3000 $ $ 2000

IL:E 1000

0 Oil? Dli2 D1'21 Di22 vi11 'ii21 \jl22

NORMAL 5000 - 2 iii q ABLATED q NORMAL EMBRYO 6 ABLATED EMBRYO !i 4000- Figure IO. The retinalvolume was calculated from areameasurements 5 ~OMTW/Al!LATEWAL of tissuesections from normaland D 1.1.1-deleted embryos that had z normal-appearingeyes. In general,the retinal volumewas smaller in the operatedembryos (p < 0.05),although most data points were within the normalrange.

Are all blastomerescompetent to form retina when located in an appropriate position? $ 1000 The deletion experiments demonstratethat many animal hemi- sphereblastomeres that normally do not produceretinal lineages 0 Dill Dl12 Dl21 D122 Vlll Vi21 Vi22 are competent to do so during the regulation that restores the retina. The transplantation experiments describedabove dem- Figure Il. A comparisonof blastomerecontribution to the retinain onstrate further that even more distant ventral blastomeresare normaland D 1.1.1-deleted embryos. Ventral blastomeres(V 1.1.1and competent to produce retina if they are moved to the appropriate V 1.2.I), which normallyproduce very few retinal cells,now produce substantialnumbers of retinal cells.Two dorsalblastomeres (D 1.1.2, position. However, not all blastomeresare competent; trans- D1.2.2)significantly decreased their contributionsto retina. plantation of dorsal- or ventral-vegetal pole cells (tier 4) into the D 1.1.1 position never resultedin labeledclones in the retina (Table 6). Instead, the labeled cells were in the gut, heart, head, and trunk muscle and nephric ducts. About 77% of these em- by a spatially confined morphogen. Equivalent studies in ver- bryos had eyes, and nearly half of thesewere smaller than nor- tebratesare difficult to design.However, becausea singleclutch mal. The tier 3 dorsal midline blastomere, D2.1.2, had small of frog embryos contains specimensin which the early cleavage clonesin the retina in only 25% of the cases.Most membersof blastomeresare nearly identical, it waspossible to test whether the clonesof the transplanted D2.1.2 were in structuresthat are there are lineage patterns that are important in the generation not typical of the D 1.1.1 position (e.g., ventral somite, nephric of retinal cells, and whether the retinal fates of blastomerepro- ducts, pharynx, and gut; Moody, 1987b). These results indicate genitors are determined by placing thesecells in novel environ- that even in the most “retinogenic” position, thesevegetal cells ments. could not or had only limited ability to produce a retinal lineage. Therefore, although positional information is important for blastomeresto expressa retinal lineage, not all blastomeresare Table 6. Percentage of embryos with labeled clones in the retina competent to respond to this information. after blastomere transplantation

Discussion Donorfrom: Hostnosition n Clonesin retina In several invertebrates it has been possibleto observe directly v1.1.2 Dl.l.l 10 100% the mitosesleading from zygote to differentiated cells. In general, v2.1.2 D1.l.l 13 100% it has been found that a cell’s genealogycan affect the deter- D1.l.l D1.l.l 9 100% mination of its phenotype in one of three patterns (Kenyon, D1.l.l v1.1.2 16 0 1985; Stent, 1985; Davidson, 1990). Cells expressingthe same v2.1.1 D1.l.l 9 0 phenotype may (1) descendfrom a common precursor, (2) share D2.1.1 D1.l.l 7 0 equivalent lineal positions within the mitotic history, or (3) be D2.1.2 D1.l.l 20 25% placed in the appropriate position by their lineageto be induced 3206 Huang and Moody - Lineage Regulation in Frog Retina

6000

A 0 NORMALD1.l.l B 0 NORMALD1.l.l q D1.l.l WITH V1.l.l ABIATED q D1.l.l WITH D1.1.2ABlATED

4000-

2 3000- kt -I- s Figure 12. Comparison of the num- j 2000- ber of labeled cells in the retina de- Y scended from blastomeres in normal 0 embryos and in embryos with various El blastomere deletions. The dots in the kz lOOO- 32-cell embryo depicted in the upper 4 right corner of each graph illustrate 1 ~ *p

Polyclonal origin of the retina Are blastomere lineagesjixed regarding the number of It is clear from previous studiesthat the Xenopus retina is de- mitoses? rived from more than one blastomereprecursor cell (Jacobson In the nematode, geneshave been identified that control the and Hirose, 1981; Dale and Slack, 1987a; Moody, 1987a,b; number of cell divisions that are characteristic of particular Moody and Kline, 1990), and it has been proposed that the lineages(Chalfie et al., 198l), and it has beenhypothesized that mouse retina arises from no fewer than 25 founder cells (Wil- similar genesin vertebrates may contribute to the regulation of liams and Goldowitz, 1992). Although the frog retina descends cell number in CNS populations (Williams and Herrup, 1985). from multiple early precursors,the specificblastomeres we iden- Although previous studiesin the retina observed a wide variety tified were invariant in the largepopulation (n = 124)of embryos of clone sizes, the randomnessin progenitor labeling did not studied, and were a restricted subsetof those blastomeresdes- allow an analysis of whether the size of a clone of a specijic tined to produce the forebrain. Our quantitation of a largenum- progenitor is invariant. In the frog, however, we had the op- ber of clones demonstratesthat the nine blastomeres(five ip- portunity to test this hypothesis by determining whether clones silateral and four contralateral) that produce the retina do not descendedfrom the sameprogenitor in many different animals all contribute the same number of cells. Based on population were the samesize. We found that the variation betweenanimals statistics,the most animal dorsalmidline cell (Dl . 1.1) produces was as much as IO-fold, considering the logarithmic nature of about half of the retina, the three other ipsilateral dorsal cells cell divisions, this is a difference of three to four mitoses.The eachproduce about an eighth of the retina, and the contralateral wide rangeof the blastomeres’contribution to the retina cannot dorsal cells and ipsilateral ventral cell together produce the re- be simply explained asthe result of the variation of the cleavage maining eighth of the retina. Thus, the founder cells are unequal plane in thesestereotypic embryos sinceprevious fate mapping contributors to this structure, a finding that contrasts the as- showed that interanimal variation is less than 10% (Moody, sumption that founder cells of the chimeric mouseretina each 1987a,b).These results indicate that blastomereprogenitors are contribute equal numbers of cells (Williams and Goldowitz, not determined to produce a fixed number of progeny in the 1992). Even though it has been noted previously that part of retina. Quantitation of clone sizesin another forebrain structure the retina descendsfrom contralateral blastomeres(e.g., Jacob- in the frog, the hypothalamic dopamine nucleus,also found 1O- son and Hirose, 1978), it was surprising that these cells ac- fold differencesin cell numbersamong clonesderived from the counted for such a large number of cells. sameprecursor (Huang and Moody, 1992). Therefore, if there The Journal of Neuroscience, Auqust 1993, U(8) 3207

Figure 13. Photomicrographsshow that the distributionof the labeleddescendants of V1.1.2 (A), which wastransplanted to the positionof Dl . 1.1, is indistinguishablefrom the patternof descendantsfrom a normalDl. 1.1 (B). mb, midbrain;Zf; infundibulum;r, retina(ipsilateral to injection).Scale bar, 200 pm. is any lineage control of the number of mitoses, it must be exerted much later than cleavage stages. 6000 Are blastomere lineagesjixed regarding spatial distribution? 5000 Previous fate maps have noted a spatial arrangement in the retina whereby dorsal blastomeresgive rise mostly to ventral i T- retina and ventral blastomeresgive rise mostly to dorsal retina 2 4000 (Jacobson and Hirose, 1978, 1981; Dale and Slack, 1987a; ii Moody, 1987a,b). In fact, Jacobson(1983, 1985) proposedthat A2 3000

clones initiated later in the blastula respect boundaries, much 8 like insect imaginal disk compartment boundaries. Such clonal G52 2000 segregationwould have an interesting correlation with the later spatial restriction of GC axons as they innervate distinctly dif- 5 1000 ferent regionsof the optic tectum. However, with precisequan- titation of the number of cells in each clone in the different retinal sectors, we found no evidence for boundaries that seg- 0 regate clones from different cleavage stageprogenitors. There 0 NORMAL Di .I .l are shallow gradients in the density of some blastomereclones q V1.1.2INPLACEOFD1.1.1 in either the anterior-posterior (nasal-temporal) or dorsal-ven- Figure 14. The numberof labeledretinal cellsthat descendedfrom tral directions, but no strict boundaries; the descendantsfrom V 1.1.2,which was transplanted to the siteof D 1.1.1,is the sameas the each blastomerecan be found in any region of the retina. Thus, number that descendedfrom normalD 1.1.1. 3208 Huang and Moody - Lineage Regulation in Frog Retina the retina is a true mosaic of the clones of several blastomere maining in the different sublineages.The size variations in late progenitors. In contrast, clones that are initiated closer to the initiated clones(Holt et al., 1988; Wetts and Fraser, 1988)prob- terminal mitosis, for example, in the retina (Holt et al., 1988; ably arise becausedifferent injected cells were in different parts Wetts and Fraser, 1988) chick spinal cord (Leber et al., 1990), of their mitotic program. This temporal mosaicismwould pre- chick hindbrain (Fraser et al., 1990), or chick optic tectum (Gray vent single cell injections from revealing a lineagemodule like et al., 1988) do show some spatial restriction. that proposed in the mouse retina (Williams and Goldowitz, 1992).Our data also do not reveal a lineagemodule that consists Are blastomere lineages fuced regarding cell phenotype? of the same number of cells. The radial clustersvaried in size In every animal studied, the clones contained all retinal cell becauseeither the founding neuroepithelial cells divide different types, indicating that blastomere origin does not dictate a single numbers of times, or the clustersdescend from more than one phenotype, and all blastomeres are able to produce multiple cell founder. types. This finding is not surprising since previous lineage stud- We do not know when the temporal mosaicis initiated during ies in both the rodent and frog retina demonstrated that retinal retinal development, but the clonal mosaic of the optic vesicle precursor cells can produce multiple cell types even at their final has its beginningsat least as early as the beginning of gastru- cell division (Holt et al., 1988; Wetts and Fraser, 1988; Turner lation, as evidenced by the intermixed blastomereclones in the et al., 1990). Interestingly, the descendants from the different anterior neural areaof the gastrulaand in the prospective retinal blastomeres are distributed in the three retinal layers in the same area of the neural plate (Figs. 7, 8). Although the lack of sharp proportion as the laminar distribution ofcells in the whole retina boundariesbetween radial subclonesin the stage43-44 retina (Table 2). However, by the time radial subclones are established, may result from a slow, progressive cell mixing (Wetts and many have a laminar distribution that is different from that of Fraser, 1989) the cell clusters in the optic vesicle are too co- the whole retina, suggesting that a laminar preference is estab- herent and are distributed too far apart in the different quadrants lished sometime during the formation of the radial subclones, of the retina (Fig. 6) for slow mixing to account for the lineage and that laminar preference is established prior to phenotype mosaic. Furthermore, the mosaic pattern observed at the be- determination. A similar conclusion has been reached for mam- ginning of eye formation is asmixed as that in the tadpole retina, malian cortical development (McConnell, 199 1). suggestingthat mixing ceasesto a large extent at the optic vesicle stage.The mosaic pattern in the retina most likely results from Formation of the columnar clone pattern extensive cell mixing during gastrulation and neurulation, which The progeny of a blastomere are not uniformly distributed dispersesthe descendantsof the blastomereprogenitors into all throughout the retina, but form small, separatedcell clusters, regionsof the presumptive retinal area;a later restriction of cell which are coherent in the radial axis. In terms of distribution movement and mixing of the descendantsof retinal neuroepi- pattern and cell composition these clustersare very similar to thelial cells resultsin the formation of columnar subclones.This those observed from labeling retinal progenitors at late devel- pattern of early lateral mixing of cells and later restricted move- opment in Xenopus (Holt et al., 1988; Wetts and Fraser, 1988) ment in the radial dimension is very similar to that reported in rat (Price et al., 1987; Turner and Cepko, 1987), and mouse the chick optic tectum (Gray et al., 1988;Gray and Sanes,199 1). (Turner et al., 1990). This fact suggeststhat at the time when theselater studies marked cells, the optic vesicle is a mosaicof Is the major progenitor necessaryfor normal retina neuroepithelial cells derived from the different blastomerepro- development? genitors. In fact, this is the case:in the optic vesicle, blastomere Although the retina is polyclonal in origin, more than half of clonesalready are dispersedthroughout all regionsof the retina the retina descendsfrom the two Dl . 1.1 cells. One classicaltest and column-like cell clustersalready are formed (Fig. 6). This of whether embryonic lineagesare irreversibly committed at a observation raisestwo questions:can our observations of blas- particular developmental time is to delete the normal progenitor tomere subclonesin the optic vesicle explain the diversity of and observewhether the tissueis restored by the remaining cells clones generated by marking cells at late stages,and how does (Davidson, 1990). For example, in tunicates if the two muscle- this mosaic retinal neuroepithelium arise? producing blastomeresare removed from the eight-cell embryo, When clonesare initiated from optic vesicle stages,they vary a musclelesslarva develops (Whittaker et al., 1977). In contrast, a lot in size (Holt et al., 1988; Wetts and Fraser, 1988). The there are many studies that show that frog embryos regulate; progenitor cells that were injected in these studies must have for example, removal of any of the eight-cell blastomeresresults been members of one of three types of radial subclones:single in a normal appearing embryo as long as one dorsal and one columnar cells, clusters of a few of thesecells, or larger clusters ventral-vegetal blastomere remains (Kageura and Yamada, containing rounded cells that no longer are attached to the lim- 1984). The present study demonstratesthat when the two major iting membranes (Fig. 6). Since retinal cells begin to leave the retinal progenitors are deleted, the remaining cells often, but cell cycle at stages24-25 (Holt et al., 1988), the rounded cells not always, reconstitute a normal-sized retina (Fig. 10). Thus, probably are postmitotic or in their final cell division, while the in the majority of cases,the Dl. 1.1 progenitor is not required columnar cells probably will go through severalmore rounds of for retinal development. This result is consistentwith previous division. Thus, if a rounded cell were labeled, its clone probably findings that spinal Rohon-Beard neurons (Jacobson,198 1 a,b) would consist of one or two cells, whereas if a columnar cell and hypothalamic dopaminergic neurons (Huang and Moody, were labeled, its clone probably would consist of several de- 1992) can be, but are not always, numerically restored after scendants.Furthermore, small groupsof columnar cells may be deletion of their major cleavage stage progenitor. We cannot a clonally related cluster that has already gone through one or ignore the fact that in many embryos the retinas were smaller two mitoses; if one of these cells were labeled, its clone would than normal. Somecases did not achieve full restoration because be intermediate in size. Thus, the Xenopusoptic vesicle appears of nonspecific damage(Table l), but in others some aspect of to be a temporal mosaic regarding the number of mitoses re- retinal fate was perturbed. The Journal of Neuroscience, August 1993, 13(E) 3209

when the more ventral-animal neighbor was removed. These Is DI. I.1 committed to a retinal fate? results suggest that deletions in the dorsal-animal region of the We were surprised that the D 1.1.1 -defective embryos had ap- embryo result in unidirectional position shifts (ventral to dorsal, parently normal eyes in the majority of cases because it has been animal to vegetal) that directly affect retinal fate. This expla- shown that the mother blastomere (Dl. 1) is determined re- nation is consistent with the results that after D 1.1.1 deletion, garding its dorsal axial fate. If grown in culture or transplanted the remaining dorsal blastomeres decrease production of retina to the ventral-vegetal pole, DI . 1 differentiates into dorsal me- whereas the ventral blastomeres increase production to become soderm (notochord) and frequently CNS (Gallagher et al., 199 1). the major progenitors (Fig. 1B). Furthermore, this movement In fact, RNA from this cell can induce ectopic dorsal axes when is consistent with the direction that the dorsal midline clones injected into ventral-vegetal cells (Hainski and Moody, 1992). take during epiboly and gastrulation (Moody, 1985; Hainski Interesting, however, is the fact that when D 1.1 or its 32-cell and Moody, 1992). Since the capacity of mesoderm to induce daughters (Gallagher et al., 199 1; present study) are transplanted neural structures and the competence of the responding ecto- to a ventral position, retinal cells are not among the progeny. derm is gradually regionalized during late gastrulation and neu- Therefore, the early commitment of this blastomere begins with rulation, and the retina-inducing capacity of mesoderm is re- regional (dorsal) rather than phenotype (retinal) specification, stricted to the middle part of the mesoderm (Saha and Grainger, as has been observed in numerous marine organisms and ver- 1992) it is reasonable that in normal embryos this mesoderm tebrates (Davidson, 1990) and in vertebrate CNS (Kimmel and is overlain mainly by the ectodermal descendants of D 1.1.1, Warga, 1986; McConnell, 199 1). and after deletion of D 1.1.1 is overlain by the ectodermal de- scendants of ventral blastomeres (mainly V 1.2.1 and V 1.1.1). Are all blastomeres competent to express a retinal fate? That position determines retinal fate is proven by the fact Deletion studies cannot address the state of commitment of the that any animal blastomere makes retina when placed in the deleted cell, but they can test whether a particular fate is re- Dl. 1.1 position. It is not known what cellular interactions con- stricted to a subset of cells. Normal fate mapping demonstrates tribute to “position” in the Xenopus embryo, although several that the retina descends from five of eight animal hemisphere types of interactions are regionalized. For example, certain pro- blastomeres per side of the 32-cell embryo, but these are not teins are synthesized (Klein and Ring, 1988) or are already the only cells capable of producing retinal cells. Two ventral present (Miyata et al., 1987) on the dorsal side of the cleavage blastomeres (V 1.1.1, V1.2.2) can become retinal progenitors stage embryo. There is a mesoderm-inducing signal specific to when D 1.1.1 is deleted, and another (Vl. 1.2) can do so when the dorsal side (Dale and Slack, 1987b) and gap junctional com- transplanted to the D 1.1.1 site. Thus, all animal hemisphere munication is more prevalent on the dorsal than ventral side blastomeres are competent to produce retinal lineages. (Guthrie, 1984). Any of these factors might provide the envi- In contrast, not all vegetal hemisphere blastomeres are com- ronment in the dorsal-animal quadrant of the embryo that ul- petent to produce retinal lineages. Transplanted V2.1.2 pro- timately leads to the determination of retinal lineages. duces retina in all cases, whereas its dorsal counterpart (D2.1.2) Nonetheless, position is not the only determinant of retinal endoderm (Moody, 1987b), produces retina when pole cells into the blastocoele at different stages that the cells transplanted to the site of the embryo that normally has the we transplanted are still multipotent (Heasman et al., 1984). greatest retina-producing potential. These differences in the Thus, their lack of competence to form retinal lineages is not competence of animal versus vegetal cells to produce retina may due to a restriction to an endodermal fate, but probably is due be related to the observations that animal pole cells remain to a lack of receptor or transduction molecule that renders their pluripotent through late blastula stages (Snape et al., 1987) descendants incapable of responding to whatever signal instructs whereas vegetal pole cells become restricted earlier (Heasman cells in the D 1.1.1 position to make retina (see Jesse11 and Mel- et al., 1984). The molecular basis of this fate restriction is not ton, 1992). Understanding the signaling and receptor molecules known, but it involves the suppression of vegetal cells from involved in the induction of retinal lineages is an important next responding to neural induction. avenue of investigation.

Blastomere position is an important determinant of retinal References fate Brun RB (198 1) The movement of the prospective eye vesicles from One interesting observation from this study is that the response the neural plate into the neural fold in Ambystomaand Xenopuslaevis. of the remaining blastomeres to the deletion of D 1.1.1 is not Dev Biol 88:192-199. Chalfie M, Horvitz HR, Sulston JE (1981) Mutations that lead to universally a compensating one; ventral blastomeres increase reiterations in the cell lineages of C. elegans. Cell 24:59-69. but dorsal ones decrease their production of retinal cells. Why Dale L, Slack JMW (1987a) Fate map for the 32-cell stage of Xenopus do different blastomeres respond differently to this deletion? It luevis.Development 99527-55 1. has long been postulated that the position of cells within a Dale L, Slack JMW (1987b) Regional specification within the meso- derm ofearly embryos ofXenopus luevis.Development 100:279-295. morphogenetic field determines that cell’s identity, with respect Davidson EH (1990) How embryos work: a comparative view of to its future fate (Wolpert, 197 1). We attempted to test whether diverse modes of cell fate specification. Development 108:365-389. changes in position correlated with changes in retinal fate by Eanleson GW. Harris WA (1990) MaDDine. of the nresumotive brain deleting cells on the animal and vegetal side of the two dorsal regions in the neural plate of &erroib lievis. J Neurobibl 21:427- midline retinal progenitors (D 1.1.1 and D 1.1.2). In both cases 440. Fraser S, Keynes R, Lumsden A (1990) Segmentation in the chick when the more dorsal-vegetal neighbor was removed, the retinal embryo hindbrain is defined by cell lineage restrictions. Nature 344: fate taken was different from its normal fate and indistinguish- 43 1435. able from that of the deleted neighbor. The same was not true Gallagher BC, Hainski AM, Moody SA (1991) Autonomous differ- 3210 Huang and Moody - Lineage Regulation in Frog Retina

entiation of dorsal axial structures from an animal cap cleavage stage teins in isolated cells of early embryos of Xenopus luevis. Cell Differ blastomere in Xenopus. Development 112: 1103-l 114. 21:47-52. Gray GE, Sanes JR (199 1) Migratory paths and phenotypic choices Moody SA (1985) Gastrulation movements of progeny derived from of clonally related cells in the avian optic tectum. Neuron 6:2 1 l-225. 16-cell stage Xenopus blastomeres. J Cell Biochem 9B275. Gray GE, Glover JC, Maiors J, Sanes JR (1988) Radial arrangement Moody SA (1987a) Fates of the blastomeres of the 16-cell stage Xen- of clonally related cellsin the chicken optic tectum: lineage analysis opus embryo. Dev Biol 119:560-578. with a recombinant retrovirus. Proc Nat1 Acad Sci USA 85:7356- Moody SA (1987b) Fates of the blastomeres of the 32-cell stage Xen- 7360. opus embryo. Dev Biol 122:300-319. Guthrie SG (1984) Patterns of junctional communication in the early Moody SA (1989) Quantitative lineage analysis of the origin of frog amphibian embryo. Nature 3 11: 149-l 5 1. primary motor and sensory neurons from cleavage stage blastomeres. Hainski AM, Moody SA (1992) Xenopus maternal RNAs from a dorsal J Neurosci 9:29 19-2930. animal blastomere induce a secondary axis in host embryos. Devel- Moody SA, Kline MJ (1990) Segregation of fate during cleavage of opment 116:347-355. frog (Xenopus luevis) blastomeres. Anat Embryo1 (Berl) 182:347-362. Heasman J, Wylie CC, Hausen P, Smith JC (1984) Fates and states Nakamura 0, Takasaki H, Nagata A (1978) Further studies of the of determination of single vegetal pole blastomeres of XenoDus luevis. prospective fates of blastomeres at the 32-cell stage of Xenopus luevis Cell 37:185-194. - - - embryos. Med Biol 56:355-360. Herrup K (1986) Cell lineage relationships in the development of the Nieuwkoou PD. Faber J (1964) Normal table of Xenonus luevis (Dau- mammalian CNS: role of cell lineage in control of cerebellar Purkinie din), 2dkd. Amsterdam: Elsevier-North Holland. - _ cell number. Dev Biol 115: 148-l 54. Price J, Turner D, Cepko C (1987) Lineage analysis in the vertebrate Holt CE, Bertsch TW, Ellis HM, Harris WA (1988) Cellular deter- nervous system by retrovirus-mediated gene transfer. Proc Nat1 Acad mination in the Xenopus retina is independent of lineage and birth- Sci USA 84: 154-l 60. date. Neuron 1: 15-26. Roux W (1888) Beitrage zur Entwicklungsmechnik des embryo. Vir- Huang S, Moody SA (1992) Does lineage determine the dopamine chows Arch Path01 Anat Physiol Klin Med 114: 113-l 53. Reprinted phenotype in the tadpole hypothalamus? A quantitative analysis. J in: Foundations of experimental embryology (Willer BH, Oppenhei- Neurosci 12:1351-1362. mer JM, eds; Laufer H, trans), pp 2-37. New York: Hafner, 1974. Jacobson M (198 1a) Rohon-Beard neuron origin from blastomeres Saha MS, Grainger RM (1992) A labile period in the determination of the 16-cell frog embryo. J Neurosci 1:9 18-922. ofthe anterior posterior axis during early neural development. Neuron Jacobson M (198 1b) Rohon-Beard neurons arise from a substitute 8:1003-1014. ancestral cell after removal of the cell from which they normally arise Snape A, Wylie CC, Smith JC, Heasman J (1987) Changes in states in the 16-cell frog embryo. J Neurosci 1:923-927. of commitment of single animal pole blastomeres of Xenopus luevis. Jacobson M (1983) Clonal organization of the central nervous system Dev Biol 119:503-510. of the frog. III. Clones stemming from individual blastomeres of the Stent G (1985) The role of cell lineage in development. Philos Trans 128-, 256-, and 512-cell stages.Dev Biol 71:191-202. R Sot Lond [Biol] 3 12:3-19. Jacobson M (1985) Clonal analvsis and cell lineage of the vertebrate Stemberg PW, Horvitz HR (1989) The combination of two intercel- central nervous system. Annu Rev Neurosci 8:7i-102. lular signaling pathways specifies three position-dependent cell fates Jacobson M (199 1) Developmental neurobiology, 3d ed. New York: in C. eleguns. Cell 58:679-693. Plenum. Sulston JE, Schierenberg E, White JG, Thomson JN (1983) The em- Jacobson M, Hirose G (1978) Origin of the retina from both sides of bryonic cell lineage of the nematode Cuenorhubditis eleguns. Dev Biol the embryonic brain: a contribution to the problem of crossing at the 100:64-l 19. optic chiasma. Science 202:637-639. Turner DL, Cepko CL (1987) A common progenitor of neurons and Jacobson M, Hirose G (198 1) Clonal organization of the central ner- glia persists in rat retina late in development. Nature 328: 13 l-l 36. vous system of the frog. II. Clones stemming from individual blas- Turner DL, Snyder EY, Cepko CL (1990) Lineage-independent de- tomeres of the 32- and-64-cell stages. J Neurosci 1:27 l-284. termination of cell type in the embryonic mouse retina. Neuron 4:833- Jesse11TM. Melton DA (1992) Diffusible factors in vertebrate em- 845. bryonic induction. Cell 68:257-270. Wetts R, Fraser SE (1988) Multipotent precursors can give rise to all Kageura H, Yamada K (1984) Pattern regulation in defect embryos major cell types of the frog retina. Science 239: 1142-l 145. of Xenopus laevis. Dev Biol 10 1:4 1o-4 15. Wetts R, Fraser SE (1989) Slow intermixing of cells during Xenopus Keller RE (1975) Vital dye mapping of the gastrula and neurula of embryogenesis contributes to consistence of the blastomere fate map. Xenopus luevis. I. Prospective area and morphogenetic movements Development 105:9-l 5. of the superficial layer. Dev Biol 42:222-24 1. Wetts R, Serbedzija GN, Fraser SE (1989) Cell lineage analysis reveals Kenyon C (1985) Cell lineage and the control of Cuenorhabditis ele- multipotent precursors in the ciliary margin of the frog retina. Dev gu:ansdevelopment. Philos Trans R Sot Lond [Biol] 3 12:2 l-38. Biol 136:254-263. Kimmel CB, Warga RM (1986) Tissue-specific cell lineages originate White JG, Horvitz HR, Sulston JE (1982) Neuron differentiation in in the aastrula of the zebrafish. Science 231:365-368. cell lineage mutants of Cuenorhubditis elegans. Nature 297:584-589. Klein SL- (1987) The first cleavage furrow demarcates the dorsal-ven- Whittaker JR, Ortolani G, Farinella-Fermzza N (1977) Autonomy of tral axis in Xenopus embryos. Dev Biol 120:299-304. acetylcholinesterase differentiation in muscle lineage cells in ascidian Klein SL, King ML (1988) Correlations between cell fate and the embryos. Dev Biol 55: 196-200. distribution of proteins that are synthesized before the midblastula Williams RW, Goldowitz D (1992) The structure of clonal and poly- transition in Xenopus. Rouxs Arch Dev Biol 197:275-28 1. clonal cell arrays in chimeric mouse retina. Proc Nat1 Acad Sci USA Leber SM, Breedlove SM, Sanes JR (1990) Lineage, arrangement, and 89:1184-1188. death of clonally related motoneurons in chick spinal cord. J Neurosci Williams RW, Hermp K (1985) The control of neuron number. Annu lo:245 l-2462. Rev Neurosci 11:423453. McConnell SK (199 1) The generation of neuronal diversity in the Wolpert L (197 1) Positional information and pattern formation. Curr central nervous system. Annu Rev Neurosci 14:269-300. Top Dev Biol 6: 183-224. Miyata S, Kageura H, Kihara HK (1987) Regional differences of pro-